Revision of Production System Rule-Bases

Murphy, Patrick M.; Pazzani, Michael J.

doi:10.1016/b978-1-55860-335-6.50032-5

Cited by 16 publications

(11 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As already noted, these are exactly the types of agents and environments that interest us, and so the concerns raised above are particularly salient for our work with interactive human-level agents. In contrast, the original CLIPS-R work (Murphy & Pazzani, 1994) examines the system's ability to correctly identify flaws in two very simple agents whose knowledge bases contain nine and fifteen rules respectively. Both agents examined in the CLIPS-R work performed tasks that were more akin to classification than they were to the highly interactive tasks that interest us.…”

Section: Model-based Diagnosismentioning

confidence: 99%

See 1 more Smart Citation

Behavior Bounding: An Efficient Method for High-Level Behavior Comparison

Wallace

2009

jair

View full text Add to dashboard Cite

In this paper, we explore methods for comparing agent behavior with human behavior to assist with validation. Our exploration begins by considering a simple method of behavior comparison. Motivated by shortcomings in this initial approach, we introduce behavior bounding, an automated model-based approach for comparing behavior that is inspired, in part, by Mitchell's Version Spaces. We show that behavior bounding can be used to compactly represent both human and agent behavior. We argue that relatively low amounts of human effort are required to build, maintain, and use the data structures that underlie behavior bounding, and we provide a theoretical basis for these arguments using notions of PAC Learnability. Next, we show empirical results indicating that this approach is effective at identifying differences in certain types of behaviors and that it performs well when compared against our initial benchmark methods. Finally, we demonstrate that behavior bounding can produce information that allows developers to identify and fix problems in an agent's behavior much more efficiently than standard debugging techniques.

show abstract

Section: Model-based Diagnosismentioning

confidence: 99%

“…In general, however, the models in these systems are relatively complicated and intended to identify problems with mechanical or solid state devices as opposed to software agents. The CLIPS-R (Murphy & Pazzani, 1994) system was designed expressly to ensure correct software agent behavior, and bears some similarity to our approach.…”

Section: Model-based Diagnosismentioning

confidence: 99%

Behavior Bounding: An Efficient Method for High-Level Behavior Comparison

Wallace

2009

jair

View full text Add to dashboard Cite

show abstract

“…Some workers have concentrated on inducing theories from scratch [15], whereas others assume we have an imperfect theory already and the examples are used for theory refinement/ theory revision (TR) [16], where an initial, flawed theory is refined. Theories may be refined or adapted [17,18], because they are inaccurate, or because they fail some operating criteria, as in the field of explanation-based learning (EBL) [19]. The TR field is also referred to as theory patching, and an early example of its use is the MIS system [20].…”

Section: Introductionmentioning

confidence: 99%

The Application of Machine Learning Tools to the Validation of an Air Traffic Control Domain Theory

West

McCluskey

2001

Int. J. Artif. Intell. Tools

View full text Add to dashboard Cite

In this paper we describe a project (IMPRESS) in which machine learning (ML) tools were created and utilised for the validation of an Air Traffic Control domain theory written in first order logic. During the project, novel techniques were devised for the automated revision of general clause form theories using training examples. These techniques were combined in an algorithm which focused in on the parts of a theory which involve ordinal sorts, and applied geometrical revision operators to repair faulty component parts. While we illustrate the feasibility of applying ML to this area, we conclude that to be effective it must be focused to the application at hand, and used in mixed-initiative mode within a tools environment. The method is illustrated with experimental results obtained during the project.

show abstract

“…Reasoning about simulated behaviour is common in planning applications [10] and so it is appropriate here for planning refinements. Refinement tools capture reasoning in different ways: CLIPS-R [7] groups traces which share initial sequences of rule firings; ODYSSEUS [11] explores all instantiations of the explicit problem solving strategy. Our problem graph and its applicability to a range of problem solving methods is described in [3].…”

Section: Representing Problem Solvingmentioning

confidence: 99%

“…ODYSSEUS [11] refines Minerva KBSs by exploring the problem-solving strategy that is explicitly represented in control knowledge. CLIPS-R [7] acts on CLIPS KBSs by analysing the KBS's interaction with the user and the content of working memory when execution halts. We developed a refinement tool specifically for the PFES shell, designed to implement formulation applications [4].…”

Section: Introductionmentioning

confidence: 99%